How a Startup Achieved Fair 24/7 Coverage Across 3 Time Zones

Overview

A fintech startup providing payment processing APIs to e-commerce platforms faced a critical challenge as their engineering team grew from 6 to 18 people across three continents. Their on-call scheduling process, originally designed for a single-office team, had become a source of constant friction. Engineers complained about unfair distribution, timezone handoffs regularly dropped incidents, and the manual spreadsheet-based scheduling consumed hours of management time weekly.

The situation reached a breaking point when two senior engineers resigned within four months, both citing on-call burden as a primary factor. With customer-facing payment processing at stake, the team needed a solution that could deliver genuinely fair 24/7 coverage without destroying team morale.

Results at a Glance

Metric	Before	After	Improvement
Shift fairness variance	3.2x difference	1.06x difference	94% improvement
Weekly scheduling overhead	4 hours	12 minutes	95% reduction
Coverage gaps per month	8-12 gaps	0 gaps	100% elimination
Weekend shift imbalance	5:1 ratio	1.1:1 ratio	78% improvement
On-call related resignations	2 in 4 months	0 in 6 months	Complete elimination
Holiday coverage conflicts	6 per quarter	0 per quarter	100% elimination

The Challenge

The startup had grown rapidly from their San Francisco headquarters. As they expanded to serve European and Asian customers, they hired engineers in London and Singapore to provide regional expertise and timezone coverage.

The 18-person engineering team was distributed across three offices: 10 engineers in San Francisco, 5 in London, and 3 in Singapore. Their payment processing infrastructure required 24/7 monitoring since transaction failures directly impacted customer revenue.

Manual Scheduling Created Unfair Distribution

The engineering manager maintained on-call schedules in a shared spreadsheet. Every two weeks, they spent four hours coordinating the next rotation, sending emails to confirm availability, and negotiating shift swaps when conflicts arose.

Despite good intentions, the manual process produced consistently unfair results. Analysis of six months of scheduling data revealed stark disparities. One San Francisco engineer had worked 14 weekend shifts in that period. Another engineer on the same team had worked only 4. The difference was not intentional but emerged from ad-hoc scheduling decisions that favored whoever responded fastest to swap requests.

“I spent my Saturday nights on call more often than not,” one engineer explained during their exit interview. “Meanwhile, some of my colleagues seemed to always find someone to cover for them. The system felt rigged even though I know it was not deliberate.”

Timezone Handoffs Dropped Incidents

The distributed team structure should have enabled follow-the-sun coverage where each region handled incidents during their local business hours. Instead, timezone transitions became danger zones.

The handoff between Singapore (end of day) and London (start of day) had a 30-minute gap where neither region had clear ownership. The handoff between London and San Francisco had similar ambiguity. During these transition periods, alerts would fire, acknowledgment would be delayed, and sometimes incidents would go unaddressed for 15-20 minutes while engineers in both regions assumed the other team was handling it.

The team documented 8-12 coverage gaps monthly, each representing potential customer impact and requiring post-incident investigation to understand what went wrong.

Holiday Coordination Was a Nightmare

Different regions observed different holidays. U.S. Thanksgiving meant reduced San Francisco coverage. UK bank holidays affected the London team. Singapore’s public holiday calendar did not align with either. The Christmas and New Year period created compounding complexity when all regions had reduced staffing.

The engineering manager spent additional hours each quarter coordinating holiday coverage, often requiring engineers to work holidays in exchange for compensatory time off. These negotiations created resentment and rarely felt fair to anyone involved.

One particularly problematic incident occurred during Lunar New Year when the Singapore team was on holiday. The San Francisco and London teams were expected to cover, but scheduling had not accounted for the extended holiday period. A payment gateway issue went unaddressed for 45 minutes while the on-call engineer in London assumed Singapore was still providing backup coverage.

The Breaking Point

When the second senior engineer resigned, explicitly stating that on-call scheduling unfairness was their primary reason for leaving, the leadership team recognized they had a systemic problem. Both departing engineers were high performers who had received competing offers but cited working conditions rather than compensation as their motivation.

The VP of Engineering committed to finding a solution that would provide verifiably fair distribution, eliminate timezone handoff gaps, and automate the scheduling overhead that consumed management time and created constant friction.

What Would Solve the Fair Coverage Problem?

Addressing the three core scheduling challenges—fairness, timezone coverage, and administrative overhead—requires multi-timezone rosters with fair distribution algorithms and automated holiday handling. Here is how these capabilities would transform the team’s on-call experience.

Regional Rosters with Fair Distribution Algorithm

Rather than maintaining a single global schedule, the solution involves creating three regional rosters, each configured with its local timezone:

Americas Roster (10 engineers): Configured for America/Los_Angeles timezone, covering 8 AM to 6 PM Pacific Time. The fair distribution algorithm tracked each engineer’s recent shift history and automatically balanced assignments to ensure even distribution of weekday and weekend coverage.

EMEA Roster (5 engineers): Configured for Europe/London timezone, covering 8 AM to 6 PM GMT. With a smaller team, the algorithm ensured that the five engineers rotated evenly through shifts rather than allowing any engineer to carry disproportionate burden.

APAC Roster (3 engineers): Configured for Asia/Singapore timezone, covering 8 AM to 6 PM SGT. The smallest regional team required careful balancing to prevent the same three engineers from feeling overwhelmed.

Each roster used the FairDistribution rotation type, which tracked historical assignments and maximized the time between each user’s shifts while ensuring balanced coverage of weekends and holidays. The algorithm considered shift history from the past 90 days when determining assignments, preventing short-term imbalances from compounding into long-term unfairness.

Coordinated Handoff Windows

The rosters were configured with overlapping coverage windows to eliminate the gaps that had plagued manual scheduling.

APAC to EMEA Handoff: APAC roster extended coverage from 8 AM to 7 PM SGT (ending at 11 AM GMT), while EMEA roster started at 7 AM GMT. This created a four-hour overlap where both regions had active coverage, ensuring smooth transition of any ongoing incidents.

EMEA to Americas Handoff: EMEA roster extended to 7 PM GMT, while Americas roster started at 7 AM PT (2 PM GMT). This provided a five-hour overlap covering the busiest period for European and American customers.

Americas to APAC Handoff: Americas roster extended to 7 PM PT (11 AM SGT next day), while APAC roster started at 8 AM SGT. This created a three-hour overlap bridging the overnight transition.

The extended coverage hours meant some engineers occasionally worked 10-hour coverage days rather than 8-hour days, but the fair distribution algorithm accounted for this by reducing their frequency in the rotation.

Automated Holiday Calendar Integration

The team integrated regional holiday calendars with each roster. U.S. federal holidays were excluded from the Americas roster, UK bank holidays from the EMEA roster, and Singapore public holidays from the APAC roster.

When a scheduled on-call engineer had a holiday exclusion, the system automatically assigned the next available engineer in the rotation. The fair distribution algorithm tracked these automatic reassignments and compensated by reducing future assignments for engineers who covered holiday gaps.

For periods when multiple regions had reduced coverage (Christmas week, New Year’s), the team established cross-regional backup rules where engineers could volunteer for holiday coverage in exchange for additional time off. These volunteer assignments were tracked separately from regular rotation to ensure they did not skew fairness calculations.

User-Specific Exclusions and Overrides

Individual engineers could set personal exclusions for vacation periods, personal commitments, or other unavailability. The system automatically advanced rotation to the next available user when exclusions applied.

The override system enabled engineers to swap shifts with colleagues when needed, but unlike the old email-based negotiation, overrides were tracked in the system and visible to the entire team. This transparency prevented the perception that some engineers were gaming the system.

When an engineer created an override to give their shift to a colleague, both the original assignment and the override were recorded. The fair distribution algorithm treated the actual worked shifts rather than originally scheduled shifts when calculating future assignments, ensuring that shift swaps did not create unfair accumulation.

What Would the Implementation Process Look Like?

A team deploying this type of scheduling system would typically complete the transition over four weeks, running parallel with their existing spreadsheet process during the transition period.

Week One: Roster Configuration

The engineering manager would create three regional rosters, configuring each with the appropriate IANA timezone identifier. Engineers would be assigned to their regional roster based on office location, with remote engineers assigned to whichever region best matched their working hours.

The fair distribution algorithm was configured with 90-day lookback for historical shift tracking. Since this was a fresh implementation, the team imported six months of historical shift data from their spreadsheets to give the algorithm context about past assignments. This ensured the new system would immediately compensate for existing imbalances rather than starting from scratch.

Week Two: Coverage Window Refinement

The team tested the handoff overlap windows by simulating shift transitions. They discovered that the initial three-hour overlap between Americas and APAC was insufficient during high-traffic periods when incidents required extended investigation. They extended the overlap to four hours by having the Americas roster end at 8 PM PT rather than 7 PM PT.

Holiday calendar integration was tested by simulating upcoming holidays. The team verified that automatic reassignment worked correctly and that the fairness algorithm properly tracked holiday coverage.

Week Three: Parallel Operation

Both the old spreadsheet system and the new automated rosters would run simultaneously. Engineers would receive notifications from both systems, allowing the team to verify that the new system’s assignments matched their expectations for fairness and coverage.

This parallel operation would reveal that the fair distribution algorithm makes different choices than manual processes. In several cases, the algorithm would assign shifts to engineers who had historically worked fewer weekends, correcting imbalances the engineering manager had not noticed in the spreadsheet data.

Week Four: Full Cutover

The team retired the spreadsheet-based scheduling and committed fully to the automated system. The engineering manager documented the override and exclusion processes, ensuring all engineers understood how to request time off and swap shifts within the new system.

Initial resistance from engineers accustomed to the informal swap negotiation process faded quickly once they experienced the transparency of the new system. Several engineers who had previously felt disadvantaged by the old process explicitly praised the visible fairness metrics.

What Results Could Teams Expect?

This type of implementation delivers measurable improvements across all three problem areas: fairness, coverage continuity, and administrative overhead.

Verifiable Shift Fairness

After three months of operation, the team analyzed shift distribution across all 18 engineers. The variance in total shifts worked had dropped from a 3.2x difference (highest to lowest) under the manual system to a 1.06x difference under fair distribution.

Weekend shift distribution showed even more dramatic improvement. Under the old system, the engineer with the most weekend shifts had worked five times more than the engineer with the fewest. Under fair distribution, the ratio dropped to 1.1:1, with the small remaining variance attributable to voluntary holiday coverage.

The fair distribution algorithm’s historical tracking meant that engineers who had been overloaded in previous months automatically received fewer assignments until their cumulative load balanced with their colleagues. This self-correcting behavior eliminated the need for manual intervention to address fairness complaints.

Zero Coverage Gaps

The coordinated handoff windows completely eliminated coverage gaps during timezone transitions. In the six months following implementation, the team documented zero incidents where alert acknowledgment was delayed due to unclear ownership between regions.

The overlap periods provided natural coordination time for ongoing incident handoffs. When an incident spanned a timezone transition, the outgoing engineer briefed the incoming engineer during the overlap window, ensuring context transfer without coverage interruption.

Engineers reported feeling more confident about their coverage boundaries. “I know exactly when my shift ends and who takes over,” one Singapore-based engineer explained. “There is no more ambiguity about whether London is awake yet.”

95% Reduction in Scheduling Overhead

The engineering manager’s weekly scheduling work dropped from four hours to approximately 12 minutes. The remaining time was spent reviewing override requests (which required approval for equity verification) and occasionally adjusting roster configurations for team changes.

The annual time savings exceeded 200 hours, equivalent to more than five weeks of full-time work. This time was redirected to technical leadership activities that had previously been deprioritized due to administrative burden.

Quarterly holiday coordination, previously a multi-hour negotiation process, became a 30-minute review of automated assignments with minor manual adjustments for volunteer coverage.

Improved Team Morale and Retention

In the six months following implementation, zero engineers cited on-call scheduling as a concern in one-on-ones or team surveys. The two positions vacated by departing engineers were filled, and both new hires reported positive on-call experiences during their first rotation cycle.

Team satisfaction surveys showed a 34% improvement in on-call-related questions compared to the pre-implementation baseline. Engineers specifically praised the transparency of the fairness metrics and the elimination of the informal negotiation that had previously favored more assertive team members.

The VP of Engineering noted that on-call discussions in team meetings shifted from complaints about fairness to constructive conversations about coverage improvement. “We went from ‘this is not fair’ to ‘how can we make handoffs even smoother’ in about three months.”

Quantified Business Impact

Beyond team morale, the operational improvements translated to measurable business outcomes. The elimination of coverage gaps reduced mean time to acknowledgment during timezone transitions by 89%. Several incidents that would have previously gone unaddressed for 15-20 minutes were now acknowledged within 2 minutes.

Customer-facing incident frequency did not change, but resolution quality improved. Post-incident reviews showed that incidents occurring during handoff periods now received the same response quality as incidents during core business hours, eliminating a previously unaddressed reliability gap.

Key Takeaways

Fair distribution algorithms eliminated perceived and actual scheduling unfairness by tracking historical assignments and automatically balancing workload across the entire team, removing the need for manual intervention or negotiation.

Multi-timezone rosters with IANA timezone support enabled each regional team to work during their local business hours while maintaining coordinated global coverage through overlapping shift windows.

Holiday calendar integration automated the complex task of regional holiday coordination, ensuring coverage continuity without requiring engineers to work holidays unless they volunteered for additional time off.

The override system provided flexibility for personal scheduling needs while maintaining transparency that prevented perception of unfair advantage, with all swaps visible to the entire team.

Importing historical shift data during implementation allowed the fair distribution algorithm to immediately address existing imbalances rather than starting from scratch, accelerating the transition to equitable scheduling.

The combination of automated fairness, timezone coordination, and administrative overhead reduction transformed on-call from a source of team friction into a sustainable operational practice that supported rather than undermined engineering culture.